Tandy Trower: Robotics Update

Sign in to queue

The Discussion

  • User profile image
    Is Microsoft doing any research with bigger/complicated robots like humanoids? How about AI in robotics?
  • User profile image
    I'd imagine we have some folks here working on AI, but the AI community has certainly gone the machine learning route, which is a long and twisted road, but of course one that is required to be able to reliably enable the type of intelligence you refer to.

    An interview is in order.

  • User profile image

    Having installed the April CLP release and working through Sara Morgan's book, I wish that Tandy had said something about the changes to the service programming model.

    Tandy emphasized the tooling issues facing the Robotics community as the driving factor in the creation of the Robotics Studio. At ICRA this year, the software working group spent a long time hashing out the problems facing the community. They were very concerned about:

    1. Real-time concerns. When I first began reading through the material on CCR and DSSP, it seemed obvious to me that there is a roadmap that should eventually address those issues. However, it hasn't been made explicit in any of the public material I've seen produced by Microsoft.
    2. Algorithmic portability. The graduate student and advisors revealed that half of a Ph.D. thesis is usually invested in cleaning up the last student's code so that it can be adapted to a new problem. This is a serious impediment to progress. Has Microsoft considered how its factoring of the implementation framework feeds into the problem of refactoring? From some discussion at the session, I believe that it has.

    If I was a serious designer working for a smaller shop, I'd like to understand how the MSRS team sees these challenges. If I was in a university, I'd like to know whether I can take my simulations and transfer them to a working robot. And, for example, I'd like to be able to understand and influence class attributes specifications to ensure that the realtime potential was not being undermined.

    This is the space I work in. How do I get that kind of information?

Add Your 2 Cents